No, study didn’t show that ‘mindfulness training doesn’t foster empathy or makes narcissists worse’

quick takes

Many of us have become accustomed to extravagant claims about the benefits of mindfulness that turn out to be based on poor quality studies with inadequate control groups. We become skeptical about what we are told about the benefits of mindfulness. We’ve come to expect a lot of confirmation bias.

Reactions to the study that I am going to be discussing, though, suggest that overexposure to these kinds of studies may create a bias of a different kind. Namely, we may have more accepting of claims of negative effects of mindfulness, even when they come a poor quality study.

A click bait headline of a link to an article in the British Psychological Society (BPS) Research Digest kept showing up in my Twitter feed.

bps on empthy study

mindfulness doesn't foster empahtyI might be inclined to believe it, without examining the evidence. Why not? It’s not at all clear that there are any specific effects of mindfulness, in the active ingredient, beyond nonspecific – placebo – conditions with which it is administered.

I suspect a lot of people who were retweeting it probably didn’t bother to check whether the article actually made sense.

When I will I obtained the open access study that had inspired the story in the BPS Research Digest, I could quickly see that the claims were not warranted.

Ridderinkhof A, de Bruin EI, Brummelman E, Bögels SM. Does mindfulness meditation increase empathy? An experiment. Self and Identity. 2017 Jan 3:1-9.

The abstract

Cultivating empathy is a presumed benefit of mindfulness, but this possibility has rarely been investigated experimentally. We examined whether a five-minute mindfulness exercise would cultivate empathy relative to two equally brief control exercises: relaxation and mind-wandering. We further examined whether mindfulness would be especially beneficial for people with autistic or narcissistic traits. Results showed no effect of mindfulness relative to both control conditions on mind reading, empathic responding, or prosocial behavior. Mindfulness effects were independent of autistic traits. Unexpectedly, people higher in autistic traits did show increased prosocial behavior across conditions. Intriguingly, mindfulness improved mind reading in non-narcissistic people, but reduced it in narcissistic people. These findings question whether a brief mindfulness exercise is sufficient for building empathy.

The study found no overall effects of mindfulness on empathy as it was measured in the study. The click bait headline was based on post hoc subgroup analyses.

When I drilled down into the article itself, I saw that it was not actually conventional mindfulness training that was provided to participants assigned to that condition, but a five-minute analog exercise.

Apparently the five-minute exercise was not very convincing to participants, because those who received it rated it as leaving them less mindful than those receiving a relaxation control manipulation rated themselves after it.

The authors nonetheless provided subgroup analyses organized around to personality variables, which they termed autistic traits and narcissism. Most of these analyses did not produce a significant effect, some were counterintuitive but the abstract and the authors discussion centered on the few of them

The authors fell into the trap of being swayed by the mere name of the measure. When is administered to a general population sample, the autism spectrum measure does not distinguish people who are more or less likely to exhibit autism spectrum characteristics. Similarly for the measure of narcissism. In each, the authors were interpreting small differences on the low end of the scale as if they were occurring at the high end.

So, the authors tackled reasonable question about whether mindfulness fosters empathy, but they did so with very weak methods. When they didn’t get positive results, they performed lots of subgroup analyses and cherry picked a few to overinterpret.

I follow some Twitter accounts because I expect them not only to alert me to findings to which I should pay attention, but because they also filter out things I should simply ignore. In this respect, the BPS Research Digest failed me. The click bait headline was simply misleading, but it did succeed in getting me and others to go to the website. The article acknowledge some of the problems of the study, but seem to dismiss them. Worse, the BPS article offered causal interpretations of what were undoubtedly cherry-picked, spurious effects.

We’re all suckers for believing effects that can be explained, even when the effects are not there.